Friday, December 29, 2017

Artificial Intelligence at the End of 2017: Three Large Questions and Answers

The idea of nonhuman intellectual capability has been around since at least the 98 years ago that author Karel Capek introduced a now-familiar word in his play R.U.R. (Rossum’s Universal Robots).  Since then, it has often gone under the name “artificial intelligence.”  As I have written before, that has been a misnomer, as everything being accomplished in it had been simply algorithmic, or limited to situations where something can be programmed to execute an “if a happens, then do b” sequence, which is not what we mean by intelligence. 

Over the past few years, though, the field has become more prominent, and its developments have reached many ordinary consumers.  People have published articles about “machine learning,” which, according to one definition, “is an application of artificial intelligence (AI) that provides systems the ability to automatically learn and improve from experience without being explicitly programmed.”  That technology has been credited for, among other things, reaching world-class status in go, a game which, along with bridge, poker, and chess, the most skilled and motivated humans can spend a lifetime learning without coming close to comprehensively solving.  We have also seen a variety of non-fundamental but large incremental improvements in home management devices and children’s friendship-robot toys.  Classic questions, such as “Will Artificial Intelligence Become Conscious?” (Subhash Kak, Live Science, December 10) and “A.I. Will Transform the Economy.  But How Much, and How Soon?” (Steve Lohr, The New York Times, November 30) have returned to the press, and salient concerns about the latest playthings have been raised by, among others, MIT professor Sherry Turkle, after decades still the leading figure on the social side of technology, whose “Why these friendly robots can’t be good friends to our kids” (The Washington Post, December 7th) was released almost simultaneously with “Should Children Form Emotional Bonds With Robots?” (Alexis C. Madrigal, The Atlantic, December). 

All of this, though, is preliminary to three higher-level questions.  Is artificial intelligence still algorithmic?  What effect, beyond the established factors of globalization and efficiency and what we have already been able to project about automation, will it have on American and world employment?  And based on where it is actually going, is it dangerous?

The first question’s answer is yes.  Getting computers to program themselves is revolutionary, and cutting out the slowest and least effective parts, the people, will prove to be a huge improvement, and has already got us unquestionably fundamental gains, with many more to follow in the next months, years, and decades.  It can also be so fast and complex that many such systems cannot explain their results in forms succinct enough for human understanding.  However, machine learning itself is still limited to computational procedures, with nothing else under the surface.

On the second issue, we have five different responses.  In the October 29th Salon “Will the AI jobs revolution bring about human revolt, too?,” Kentaro Toyama saw more to artificial intelligence than actually there, and interpreted author Ray Kurzweil’s old Singularity prediction as meaning that “by 2045, computer intelligence will match or exceed human abilities in every way.”  While it may be true that even now a product can come up with “eerie, dream-like images that seem genuinely creative and uncomfortably human,” that does not mean that such things are truly either, and employment requiring authenticity there will not go away.  In ”Who’s afraid of the A.I. wolf?” in the same publication seven days later, Crystal Point gave us the opposite view, saying correctly that “scientists still haven’t pinpointed the actual brain processes that make up awareness, and philosophers have not reached a consensus on the nature of this baffling state.”  In the November 9th Motley Fool, Chris Neiger went back to the first position in “Artificial Intelligence Is Already Common – and It’s About to Take Over,” naming a study claiming that 38% of American jobs could go away to automation by 2030, a valid prediction in any event, and that artificial intelligence could “be capable of performing any task currently done by humans,” if not up and running in all of them, by 2060.  Lohr’s article above presented both sides, and a set of specific timelines, with 25% and 75% chances of success polled from “hundreds of attendees at two well-regarded AI conferences,” turned up in “Human obsolescence” in The Economist’s December The World in 2018 issue.  A few of the capabilities rated included “fold laundry” (25th percentile of probability 2018, 50th in 2022, 75th in 2031), “retail salesperson” (25th in 2022, 50th in 2029, and 75th in 2048), ending with “full automation of labor” (25th percentile 2071, 50th percentile 2141, and 75th percentile 2241).  Despite the conference connection these chances also include existing forms of automation, but are enough to tell us that even those in the field do not think any kind of “singularity” is likely within anything like 28 years. 

On the third question, it is still a real concern that such machines will program themselves with systems that take living beings to be unnecessary, redundant, or even detrimental to their purposes.  The movie The Terminator may now be 33 years old, but its premise of “autonomous goal-seeking programs” failing to incorporate basic human assumptions is as current as today’s news.  Those involved with artificial intelligence must take a line from another 1980s popular culture work, Donald Fagen’s song “I.G.Y.,” and ensure that if we have “a just machine to make big decisions,” it is “programmed by fellows with compassion and vision.”  That is true whatever the eventual effects of artificial intelligence.


Friday, December 22, 2017

Driverless Cars – Wrap-Up, The New York Times Magazine Issue, and Ten Things to Understand

Since and including October 13th, when I said I could write a weekly blog on this topic alone, I have done almost that.  Except for two monthly issues on the AJSN and the jobs numbers, this source has provided nothing but eight topical installments.  Now, finally, it’s time to summarize, in preparation for moving on to other issues, this jobs-and-way-beyond area.

First, here is a glancing mention of nine remaining articles.  In “Waymo’s Human Problem” (Forbes, November 8th), Chinka Mui addressed yet another quandary, that of how to deal with shared-driverless-car users leaving behind a mess, which, as any current or former cabdriver (present company included) will tell you, is hardly rare.  We learned that “Optimus Ride will provide self-driving vehicles to Boston community residents” (Darrell Etherington, TechCrunch, November 30), another example of the advantage of introducing autonomous cars and buses in agreeable subdivisions.  The tech-happy Russians are showing their skill in “Yandex takes its self-driving test cars out for a spin in the snow” (Natasha Lomas, TechCrunch, November 28th) – where could anyone better learn about that?  Unsurprisingly, “Intel wants to make your autonomous car rides more entertaining” (Andrew Tarantole, Engadget, November 29th), with “in-cabin, immersive experiences,” which will be a great growth area within this larger one.  We can expect “GM to launch self-driving vehicles in big US cities in 2019” (Fox Business, November 30th), if that company, which seems behind the curve, can get there.  We saw more specificity by one of the leaders in “Ford details plans for all-new autonomous vehicle” (Jeff Flock, Fox Business, December 6th), which will, quite appropriately in my view, be “shifting its focus from a new electric car driven by traditional buyers to hybrids driven by no one at all,” and will strive to “enable the vehicles to be in constant use.”  Central Beantown’s chaotic driving will improve, as “Lyft’s self-driving pilot with nuTonomy begins rolling out in Boston” (Matthew Lynley, TechCrunch, December 6th), with that firm seeming soberer and more measured than Uber.  The chance of a large computer internals maker, called “Chipzilla” without irony here, to appear in competing consortia now seems strong, as described in “Mobileye’s Latest Moves Will Strengthen Intel’s Clout in Autonomous Cars” (Motley Fool in Fox Business, December 9th), and, as to another behemoth, “Apple AI chief reveals more progress on self-driving car tech” (Jon Fingas, Engadget, December 10th).  Christina Anderson and Neal E. Boudette wrote that Volvo, the most famous-for-safety automaker, is “Trying to Bypass Anxiety on the Road to Driverless Cars” (The New York Times, December 12th), a fine thing for the industry, and one reason why political-liberal concern has significantly dropped in the past few months.

Second, I have not reviewed the year’s largest event in autonomous-vehicle press, the November 12th New York Times Magazine number, titled “Life After Driving.”  The entire section, except for the ads and puzzles, was given over to five articles on driverless cars, some with multiple parts, on what is happening, what will probably happen, what may or may not happen, and their myriad potential effects on our lives.  Given that lead times for books are simply too long, this remarkably current 82-page compendium will need to stand in, and it does that superbly.  The Sunday Times may cost $7 per issue here, but this thoughtful and prophetic compendium alone, which ended with a full-page drawing of “the museum of driving,” at which our descendants will be able to experience, among other things, drag racing, sitting in traffic jams, pumping gas, parallel parking, and even finding their cars in parking lots, was easily worth three times that.  If you haven’t seen it, it is well worth getting a copy.

Third, here are ten things we should all understand about autonomous vehicles and their associated revolution.

Number one, despite many efforts to tether the two together, driverless does not mean electric.  Hybrids would be well-suited, but after 50-plus years of government promotion and optimistic forecasts, only 2% of United States cars are electric-only, and that share, even if it goes up, will have its figurative doors blown in early next decade by the share of autonomous ones.

Number two, rural and urban considerations will be different.  Here in the rural Catskills, it will not be expedient for me to take a shared car one mile to the town center, 15 miles to the nearest McDonald’s, or 35 to the closest major retail center.  On the other hand, most city residents will join those in Manhattan by finding that owning a car is unjustified.

Number three, expected job losses are no reason to stop or deter technological progress, even if thoughts that other positions will replace them are unjustified.

Number four, in life quality and even expectancy, autonomous vehicles will prove to be a great boon.  Half a century from now, we will wonder how we lived without them.

Number five, consumer acceptance of driverless cars will grow, though not necessarily quickly.  Once people ride in them, see they are not scary, and understand their rapid improvement pace, few will stand against their proliferation.

Number six, the effects will be extremely broad-based.  Did Gottlieb Daimler know that by putting a lawnmower-sized engine on a four-wheeled bicycle, he would be kicking off changes ranging from romance and sex practices to city layouts? 

Number seven, despite some wonderfully creative thinking, autonomous vehicles will have many consequences which we cannot now predict.  We just can’t see it all from here.

Number eight, the cars’ interiors will have a gigantic range of possibilities.  They will provide some badly needed diversity in auto design, and it is quite likely that customers will be able to choose between ones made up as offices, shopping malls, reading spaces, card clubs, or even, of course, bedrooms.

Number nine, the 2020s will be the transitional decade during which drivered and driverless vehicles, both in great numbers, will share the roads.  The challenges of minimizing accidents and maximizing efficiency will be greatest then, and, eventually, most Americans will look forward to the almost-driverless-only 2030s and beyond.


Number ten, it IS happening.  It will not turn out like artificial hearts or air-and-road vehicles, which, although they exist in niches now, failed to gain widespread use.  It will be more like email or the Internet, with autonomous cars permanently entering one life after another.  That is the most important point here.  We all need to accept that cars will soon be driverless, so we can best deal with their expected, unexpected, and potential shortcomings.  We have no other choice.

Friday, December 15, 2017

Driverless Cars – More Progress and Positioning – VIII

Another good move by the industry leader came forth in “Waymo inches closer to driverless car launch with repair deal” (Steve Dent, Engadget, November 2nd).  The partner is AutoNation, cited as “America’s largest auto retailer,” which will use its skill at maintaining massive numbers of cars to meet the expectation that, as their CEO put it, the autos will “be in service for hundreds of thousands of miles, much more than personal-use vehicles, to make them economically viable.”  Will they or won’t they?  Their chances would be best if they were diesel, but we know by now that won’t be the case.

Pedestrians can usually cross streets wherever they want, laws or not, so it would be mostly a formality if “Self-driving cars could make jaywalking legal” (Matt McFarland, cnn.com, November 3rd).  This “crime,” which the article points out was invented with the coming of cars with drivers, may go out with their replacements, which will be able to stop more consistently.  In any event, it would be good riddance to lose this infraction so small that few police even care about it.

In Gizmodo on November 3rd, Kate Conger addressed an attitude change in “We Need To Be Okay With Self-Driving Cars That Crash, Researchers Say.”  She rightfully contrasted the attention given Joshua Brown’s death, which involved driverless technology, with 37,000 annual others which did not, and publicized someone else’s question of just how safe autonomous vehicles need to be for them to be legal, concluding that if they are “just a little bit” less dangerous, they could save massive numbers of lives.  It’s now clear, though, that there will be no great cutover, but phased in a city and state at a time, which will provide plenty of data encouraging other places to accept them.  Conger’s view was more or less echoed by Aarian Marshall in Wired’s November 7th “To Save the Most Lives, Deploy (Imperfect) Self-Driving Cars ASAP,” in which the author points out that “today, autonomous vehicles are about as good as a standard crappy driver.”  Of course, though, unlike the latter they will improve, month by month.   

Michelle Chavez, writing in Fox News on the same date, told us about more progress from a beleaguered company in “Uber prepares next generation of self-driving cars.”  They may be overstating a bit, though, to say that their vehicles “could be on the road as early as the end of this year,” with so little time remaining, but it still looks like a good development for a firm that badly needs to get rid of its human drivers.  That company started off two more promising headlines later, in CNBC’s November 20th “Uber has an idea to keep you from getting sick when you read in self-driving cars” (seasickness, which is genetic, may well need attention here), and “Uber Strikes Deal With Volvo To Bring Self-Driving Cars to Its Network” (Mike Isaac in The New York Times among many other sources, also November 20th).  On November 7th, Megan Rose Dickey’s TechCrunch “Renault shows off self-driving car that can avoid obstacles as well as pro test drivers” highlighted more real improvements.   

Three days later, though, wasn’t too late to see that “Las Vegas expands its self-driving shuttle tests this week” (Jon Fingas, Engadget) – the full-year trial was scheduled to start November 8th, and did.  The debut made the news, but not for anything good, as within an hour the bus collided, though at only a couple of miles per hour, with a truck that backed improperly, getting its driver a ticket.  Jeff Zurschmeide, who got his account, “I was on the self-driving bus that crashed in Las Vegas. Here’s what really happened,” published by Digital Trends on November 9th, called it “the result of human error,” but pointed out that the truck driver may have been misled, as the shuttle did not stop and give him more space, or even sound its horn.  Is this another situation calling for different programming, or just one telling us that we need to realize that autonomous vehicles will, indeed, drive with great caution but no imagination?  It’s a good real-life case to study. 

In Forbes, Bernard Marr took a loftier view in the November 6th “The Future Of The Transport Industry – IoT, Big Data, AI and Autonomous Vehicles.”  He offered two reasonable predictions for 2020, that there will be 10 million driverless autos sharing the roads with 250 million “smart cars,” a phrase he uses not to describe tiny German vehicles but “cars connected to high-tech networks.”  We will depend on that grid, which is in its infancy now, to get us through the rough times when large numbers of both meatmobiles and self-driving vehicles will be out there.  

A worthwhile summary of an area addressed little in the popular print was Yahoo Finance’s November 9th “The ‘Driverless’ Car Era:  Liability Considerations.”  We don’t know who or what will pay when autonomous vehicles do damage from negligence, but we will resolve that issue somehow.  Meanwhile, Detroit Free Press paid attention to even more difficult issues with Todd Spangler’s November 23rd “Self-driving cars programmed to decide who dies in a crash.”  Disasters will happen, people will argue about them, and ethicists and philosophers are starting to weigh in.  It won’t be easy for them.  As computer programming demanded that we be specific about our instructions, driverless vehicles will require that we be explicit, as well as standardized, about our ethics.  In the over 2,500 years this field has been around, we haven’t come close to that – but we will, ultimately, have no alternative.

We finish today’s post, and the last regular installment of this series, with three positive pieces.  In the November 12th New York Times “Where Self-Driving Cars Go to Learn,” Cecilia Kang showed us how Arizona’s relatively laissez-faire legal stance has helped the state as well as the industry.  Washington Post’s “Driverless cars may help disabled, elderly,” reprinted in the November 24th Times Herald-Record, asked questions about wheelchair accessibility, which, if required for all autonomous taxis, could massively increase their cost and even their sizes, but acknowledged that such vehicles, even with less expensive accommodations, will be flexible and greatly beneficial for people with other disabilities.  Joe Rinzel got “Driverless cars can transport lives – if we change the rules and let them” on the November 21st USA Today editorial page – the article, as well as announcing that Central Florida has joined Michigan, California, and the state above in the front line of the technology, made a case for friendlier laws nationwide. 


Next week, I will address the New York Times November 12th magazine issue, which was dedicated to the subject of this series, and wrap up, for now, with some conclusions.  While I now need to allow room for other subjects, there will be much more here on driverless cars in the years to come.  

Friday, December 8, 2017

November’s Employment Data: A Good Jobs Gain, Few Other Changes, AJSN Effectively Identical

This may be the closest our nation’s employment situation has ever come to being unchanged by a Bureau of Labor Statistics Employment Situation Summary.

The good news, and almost all the news for that matter, was the creation of 228,000 net new nonfarm positions, almost 100,000 more than population growth and about what some analysts projected.  Otherwise, little happened.  Adjusted and unadjusted unemployment were unchanged at 4.1% and 3.9% respectively.  There are still 1.6 million people officially jobless and without work for 27 weeks or longer.  The labor force participation rate held at 62.7%, and the number of Americans working part-time for economic reasons, or holding on to less than full-time propositions while looking beyond that level, stayed at 4.8 million.  Average private nonfarm payroll earnings, after a downward 3-cent October adjustment, went up 5 cents per hour to $26.55, the closest rounding-to-the-nearest-penny figure to inflation.  Two major metrics had one-increment worsenings, as the employment-population ratio decreased 0.1% to 60.1% and the number of unemployed went up 100,000 to 6.6 million.

The American Job Shortage Number or AJSN, the Royal Flush Press statistic showing how many new positions could be quickly filled if getting one were known to be easy, got into the breakeven act with distinction, down from October’s 16,169,554 to 16,169,313 for an improvement of 242, or .0015%.  That, while humorous, is of course statistically insignificant – in fact, the same would be true for a gain or loss of 100 times that amount – so we can safely say the metric in November was unchanged.  For the sake of completeness, here is the latest breakdown:


The latent demand effect of none of the above categories changed more than 49,500, that coming from a drop in the number of discouraged workers from 524,000 to 469,000.  The share of jobs that would go to those officially unemployed is still a hair short of 35%. 

There were greater differences between last month and November 2016.  The largest change in our job shortage came from official unemployment’s fall from 7,066,000, which, with a latent demand drop of 702,000, was very close to the AJSN’s 714,000 one-year decline. 


Was November, then, good?  The best thing about it, aside from the gain in jobs, is that we are camping at some good levels – those unemployment rates, the decline in those out for half a year or more, and undistinguished but off-recent-low working and participation rates.  Worst was that we really went nowhere, meaning our low jobless figures are still heavily dependent on people leaving the labor force.  Our monthly job gain did not seem to help the other numbers at all.  The turtle, then, stayed right where he was.  

Friday, December 1, 2017

Driverless Cars – More Progress and Positioning – VII

An American government organization wanting, apparently by its own choice, to regulate less!  That’s the subject of “NHTSA seeks to remove old obstacles to clear the way for self-driving cars” (Eric Brackett, Yahoo News, October 28th).  
Yes indeed, the National Highway Traffic Safety Administration “was seeking input on how it could remove regulations that are slowing down the production and deployment of self-driving cars.”  These rules haven’t stopped much so far, but if unchanged they will soon.  It’s time for discussion, negotiation, and official recognition that the nature of vehicles is changing, and hats off to this rare federal agency for seeing that.  Meanwhile, in the same source and on the same date, per author Trevor Mogg, “Waymo tootles into Detroit with its self-driving car project,” where, at its proving ground actually in Novi, it will learn more about driverless ice and snow coping.  

It’s better for any vehicle to make its mistakes in cyberspace instead of in bricks-and-mortarland.  On October 29th in The New York Times’s “What Virtual Reality Can Teach a Driverless Car,” Cade Metz recapped the state of this art, including how much faster such “learning” can be at computer speeds, the “complete control” researchers can use, and the problems of checking up on what machines are ready to implement.  The best use for this tool seems to be along with human perceptions and direction, especially in developing subsystems, such as determining braking speed and intensity, where theory may be insufficient. 

There are many reasons why first-world cities will adopt autonomous vehicles the earliest.  One is how the humans drive there.  In Wired.com’s “Prepping Self-Driving Cars for the World’s Most Chaotic Cities,” also on October 29th, Kaveh Waddell contrasted the most disordered American cities, properly using Boston as an example, with those in “developing countries” with “huge, anarchic intersections” and “drivers who have little to zero respect for lanes, traffic signals, warning signs, and speed limits.”  There is a gap between programming the possibility of vehicles going through red lights a second or two late and dealing with those disregarding all laws.  There is also a difference between stricter enforcement of traffic regulations all drivers understand and the need to create order where there is none.  This problem will keep much of the world’s autonomous vehicle saturation and even significant introduction well behind.

China presents its own problems, such as the previous article’s note that different regions there have sharply differing traffic signs and customs.  Yet its own driverless company is at least trying to forge ahead.  John Fingas’s Yahoo Finance “Baidu teams with ride-hailing service to fast track self-driving cars,” still from October 29th, told us how that firm is working with national ride-hailing company Shouqi and its extensive mapping knowledge to offer self-driving vehicles themselves along with trips in them.  In the United States, “Toyota will test autonomous cars at California’s GoMentum station” (Darrell Etherington, also Yahoo Finance, October 30th), that location a proving ground in Walnut Creek.  Related activity from the same source on this paragraph’s third continent is planned for December 4th and 5th, featuring the CEO of one European startup.  To learn more about what’s planned for this conference, see “The race for the autonomous car is on, and hear Five AI attack plan at Disrupt Berlin,” by Mike Butcher, also on October 30th.

The farthest-reaching October 30th driverless vehicle story, though, came from Wired.  In “How to Design Streets for Humans – And Self-Driving Cars,” Aarian Marshall took an urban-planning view on how cities might change once human vehicle operators go away, using a National Association of City Transportation Officials’ “50-page blueprint” – for example, with thinner travel lanes, “tiny parks” instead of parking meters, crosswalk removal, paved areas used for both rush-hour travel lanes and delivery vans, and “at night, street space next to bars could be dedicated to picking up and dropping off carousers from driverless taxicabs.”  All are reasonable possibilities.

Consumer Reports, a source of hitherto unbridled skepticism about self-driving technology, looked “Inside Waymo’s Self-Driving Car Castle” (October 31), and found “a mock community” with that company’s employees, termed ““Fauxes,” who ride bikes, jaywalk and drive cars erratically in a bid to get the self-driving software to understand how to drive in the real world.”  The article described excellent results coping with such things as “a Faux with car trouble… walking around a disabled vehicle holding his head, talking on a cell phone” and “a group of four Fauxes pretend(ing) to be a sloppy moving crew,” who “spill boxes out onto the road in front of a Waymo test car.”  This work is one of the exact things on which driverless vehicle consortia need to focus, and their ability to deal with these problems, already, should ease concerns of Consumer Reports readers and others not believing they can do that.  Related, “Waymo’s CEO says self-driving cars are ‘really close’ to being ready for the road – but plenty of challenges remain” (Troy Wolverton, Yahoo Finance, also October 31st.)  This company leader, John Krafcik, talked forthrightly with reporters about its strengths and current weaknesses, the latter including a self-driving vehicle, that if faced with “a moving van that was double-parked,” might never move without human intervention.  On the other Halloween-published issue about that company, “Waymo’s self-driving car challenge:  Making it easier to pick up passengers” in CNBC, the solution seems simple:  Ask a taxi driver!  Cabdrivers become experts at judging where their customers will get into and out of their vehicles, and most if not all of that thinking can be quantified and programmed.

We end Installment 7 with a famous author and “avowed car buff” who would do well to read this blog.  When “Malcolm Gladwell looks at the future of self-driving cars” (CBS News, once more October 31st), he apparently sees things that have been in the press for months if not years:  that “a host of issues must be resolved before self-driving cars hit the streets en masse,” the need for “a social calculation” when deciding what to collide with, the hacking problem, and losing “the pleasure that many people get from driving.”  These Gladwell might be publishing now, if he had just released a book with content put to bed six months or more ago.  Yet developments in this field are too rapid for that; a book on the subject, even if researched and written impeccably, would be obsolete on release.  Accordingly, let’s hope that Gladwell doesn’t write one soon.